Power Supply -12v question

  • Thread starter Thread starter Bob
  • Start date Start date
B

Bob

My P3BF equipped system is halting on boot and displaying an error
on the -12v monitor showing only .5v. Everything else looks normal.
Bypassing this error lets it boot and run normally.

I'd like to test the PS instead of just yanking it but I am not
familiar with how these are wired. Which wire in the supply harness
is -12v ? Is this a separate tap off the transfomer/rectifier or just
the ground side of power... meaning I have a really strange problem
here ? Why does it run normally with this -12v deficiency ?

Thanks for any insight.
 
Bob said:
My P3BF equipped system is halting on boot and displaying an error
on the -12v monitor showing only .5v. Everything else looks normal.
Bypassing this error lets it boot and run normally.

I'd like to test the PS instead of just yanking it but I am not
familiar with how these are wired. Which wire in the supply harness
is -12v ? Is this a separate tap off the transfomer/rectifier or just
the ground side of power... meaning I have a really strange problem
here ? Why does it run normally with this -12v deficiency ?

Thanks for any insight.
Click to expand...

http://www.formfactors.org/FFDetail.asp?FFID=1&CatID=1
http://www.formfactors.org/developer/specs/atx2_1.pdf (pg.13)

Pin 1---> <---Pin 11 (Looking at end of PS cable)
+3.3VDC x x +3.3VDC
+3.3VDC x x -12VDC
COM x x COM
+5VDC x x PS_ON#
COM x x COM
+5VDC x x COM
COM x x COM
PWR_OK x x -5VDC
+5VSB x x +5VDC
+12VDC x x +5VDC

All the outputs are referenced to "COM". COM is connected to safety
ground and the chassis, so COM is effectively GND as well. The
PS has one primary side (switched DC) drive, and the secondaries
are all separate windings that get their energy from the same
transformer core. The secondaries are all connected to COM.

----------
|--- +12
|--- +5
|--- +5VSB (standby supply)
|--- +3.3
|--- COM
|--- -5
|--- -12
----------

If the -12V is reading +0.5V, that sounds like it may want to go even
higher positive, but a diode is keeping it clamped close to ground.
It could be that some other positive supply rail is shorted to it,
or is partially shorted. That could be through the motherboard
itself, or it could be something gone wrong inside the PS
(don't open the PS, as there is great danger in there!).

One purpose for -12V, is to supply power to an RS-232 serial interface
chip. Since there are RS-232 translator chips that actually make their
own +/- supplies from a +5V input, this isn't absolutely necessary any
more, and whether it is needed or not, depends on how old the
motherboard is. In fact, the reason some -12V outputs read so high
in the hardware monitor, is because no current is being drawn from
the output, and the supply tends to float on the high side as a result.

The -5 similarly is little used, being suited at one time for
running some ECL circuitry or perhaps something like a Brooktree
video DAC on an old video card.

Many modern PS pay lip service to the -5 and -12V, by having very
low output current capability on these outputs.

You can try flipping the motherboard up on edge (or have someone
hold it for you), while you probe the -12V and COM pins with a
voltmeter. Or maybe do a PS swap, just to see if the symptoms
are still the same. Considering the reliability of power supplies,
you might find a different supply will fix it. A less likely cause
will be a failure inside the motherboard. A third cause could be
a poorly placed brass standoff shorting to the bottom of the
motherboard.

HTH,
Paul
 
You can try flipping the motherboard up on edge (or have someone
hold it for you), while you probe the -12V and COM pins with a
voltmeter. Or maybe do a PS swap, just to see if the symptoms
are still the same. Considering the reliability of power supplies,
you might find a different supply will fix it. A less likely cause
will be a failure inside the motherboard. A third cause could be
a poorly placed brass standoff shorting to the bottom of the
motherboard.

HTH,
Paul
Click to expand...

Thanks Paul. Is the -12v just the "bottom" of the sine wave off
the rectifier ? (I'm trying to get a conceptual handle on this).
SO, when I test with a polarity oriented meter, is it still going
to be +12v effectively between the -12v lead and the ground/common ?
Or, I am misunderstanding how they get -12v from a PS ?

Thanks,
Bob
 
It's a switching power supply, so your terminology isn't quite correct,
but the -12 volt supply is -12 volts with respect to ground (the chassis
of the power supply), while the +12 volts is at +12 volts with respect
to ground. If you measured from -12 to +12 volts, you would read 24
volts. In both cases, the absolute value, measured from ground, is 12
volts. Internally, the -12 volts is PROBABLY developed from a separate
winding on the switching power supply pulse transformer. There is very
little current available at that voltage, it's only use is for some
"old" parts and, in particular, for RS-232 drivers.
 
Bob said:
Thanks Paul. Is the -12v just the "bottom" of the sine wave off
the rectifier ? (I'm trying to get a conceptual handle on this).
SO, when I test with a polarity oriented meter, is it still going
to be +12v effectively between the -12v lead and the ground/common ?
Or, I am misunderstanding how they get -12v from a PS ?

Thanks,
Bob
Click to expand...

Consider it to be -12V DC. if you have an old analog meter, set it
to DC volts (on my old analog, this would be the 30VDC scale). Since
my old analog only measures positive voltages, connect the red lead to COM
and the black lead to the -12V pin, then reverse the sign of the
voltage you read off the meter. There should be an undetectable
amount of ripple on the signal, so it isn't going to show up on the
meter. The switching supply operates at a couple hundred kilohertz,
so there is no 60Hz involved in the outputs.

If you are curious, this web page contains a schematic traced by
hand (reverse engineered) from an ATX power supply. The parts that
makes the supply dangerous are the two electrolytic caps (C5 and
C6) that are charged to line voltage by a bridge rectifier. This
smooth high voltage DC is what is pulsed into the transformer at
a couple hundred kilohertz. The reason for doing it this way, is
the transformer has excellent coupling without requiring a huge
chunk of metal for the core. To do the equivalent at 60Hz would
require a transformer the size of a Geo Metro.

http://www.pavouk.comp.cz/hw/en_atxps.html

HTH,
Paul
 
Bob said:
My P3BF equipped system is halting on boot and displaying an error
on the -12v monitor showing only .5v. Everything else looks normal.
Bypassing this error lets it boot and run normally.

I'd like to test the PS instead of just yanking it but I am not
familiar with how these are wired. Which wire in the supply harness
is -12v ? Is this a separate tap off the transfomer/rectifier or just
the ground side of power... meaning I have a really strange problem
here ? Why does it run normally with this -12v deficiency ?

Thanks for any insight.
Click to expand...

The easiest way to check the actual voltage from the PSU is to poke your
meter probes into the motherboard connector beside the wires - black is
ground, blue is -12v - while the system is running. The meter should
read -12v DC with respect to ground.

AFAIK the only thing that won't work without a -12v supply is the serial
(COM) ports, and even then many serial devices (e.g. mice) convert RS232
levels to TTL so the ports may appear to work normally anyway.

The BIOS voltage displays come from a Winbond hardware monitoring chip,
but it only tolerates inputs in the 0 to +4v range, so there's an opamp
and resistor network to invert and scale the -12v rail between the power
supply connector and the monitoring chip. If the power supply measures
-12v but the BIOS reports 0.5v, the opamp has likely failed.

P2B
 
To output -12 volts ... AC mains voltage is converted to
300+ DC. That in turn is converted to radio wave voltage at
something like 20 kilohertz. Then through an isolation
transformer to lower voltage to the 3.3, 5, and 12 volt
ranges. Then rectified (converted to DC) again, then
filtered. Long way from AC mains to DC output which is why
computer power supplies are so resilient compared to other
electronics.

Procedure to verify voltages and how a power supply system
works is summarized in "Computer doesn't start at all" in
alt.comp.hardware on 10 Jan 2004 or
http://tinyurl.com/2t69q
Voltage limits and wire color:
Voltage Wire Color Min V Max V
+5 V Red 4.75 V 5.25 V
-5 V White -4.75 V -5.25 V
+12 V Yellow 11.4 V 12.6 V
-12 V Blue -11.4 V -12.6 V
+3.3 V Orange 3.135 V 3.465 V
+5VSB Purple 4.75 5.25
!Power On Green 0.8 2.0
Power OK Gray >2.4 when power is good

Is that -12 volt monitor from the BIOS? If so, not valid
for actual voltage numbers. Only reliable for detecting
voltage variations. As that procedure required and as
necessary: 3.5 digit multimeter. There is no faster, cheaper,
and sufficient method without even more expensive equipment
such as an oscilloscope.

Yanking a power supply takes longer, typically costs more,
is not reliable, and is what scam car mechanics do when they
too have 'no idea'.
 
Testing must be of the 'system' that includes power switch
and control electronics on motherboard. Furthermore, without
the load of a computer, other possible power supply problems
will not be obvious. Just two more reasons why better
information is available immediately (in minutes) without
removing or swapping anything. That 3.5 digit multimeter
essential (minimally necessary tool) to identify and solve
your problem. Procedure provided in that other post.

Voltages must appear in the upper 3/4 of those limits.
Limits being why analog meters are typically not sufficient.
Price being why multimeters are a tool as ubiquitous in
computer work as a screwdriver.
 
The BIOS voltage displays come from a Winbond hardware monitoring chip,
but it only tolerates inputs in the 0 to +4v range, so there's an opamp
and resistor network to invert and scale the -12v rail between the power
supply connector and the monitoring chip. If the power supply measures
-12v but the BIOS reports 0.5v, the opamp has likely failed.
Click to expand...


OK... I tested the PS at the connector. I'm getting -12v at the
connector so I don't think it's the PS. Where do I test from here ?
This is a P3BF mobo. My goal is to get it to boot without halting
on a hardware error... I don't really care is it's a hack to
repair it as I don't use the RS232 so a workaround is OK.

Thanks,
Bob
 
Bob said:
OK boys, I'm starting to clue up. If I'm reading this correctly,
the PS will have to be connected to a mobo and powered on to
pull readings from it, right ? Otherwise it doesn't charge ?
Click to expand...

Well, in the past I've been recommending that the PS be disconnected
and tested separately. But, recently, someone questioned this
practice, and it is better to keep the PS loaded while testing it.
At the very least, some supplies have minimum load ratings for some
of the rails, and the outputs won't give correct voltages without
some kind of load on them. I made myself a set of power resistors
to use as a load for the outputs (drawing a few amps from the strongest
rails and about half an amp from the weaker ones). Without a set
of dummy loads, the motherboard is the next best thing.

In fact, you can disconnect the PS, and it can be turned on via
connecting PS_ON# to an adjacent COM pin. But, doing so really
doesn't prove anything, other than that the fan spins. Having a
proper load should allow the power supply to regulate properly.

HTH,
Paul
 
Bob said:
OK... I tested the PS at the connector. I'm getting -12v at the
connector so I don't think it's the PS. Where do I test from here ?
This is a P3BF mobo. My goal is to get it to boot without halting
on a hardware error... I don't really care is it's a hack to
repair it as I don't use the RS232 so a workaround is OK.

Thanks,
Bob
Click to expand...

Since the PS is supplying the correct voltage, the RS232 ports should be
operating normally - all you have is a monitoring error.

You should be able to achieve your goal by simply configuring the BIOS
to ignore the erroneous -12v reading, it will then boot without
complaining about a hardware error.

If you want to fix the monitoring error, you'll need to trace and test
the circuitry between the PS connector and the monitoring chip - I'm not
familiar with the details in this area of the P3B-F, so can't be more
specific.

P2B
 
You should be able to achieve your goal by simply configuring the BIOS
to ignore the erroneous -12v reading, it will then boot without
complaining about a hardware error.
Click to expand...

Unfortunately there's no "ignore" feature for the voltage monitoring.
If you want to fix the monitoring error, you'll need to trace and test
the circuitry between the PS connector and the monitoring chip - I'm not
familiar with the details in this area of the P3B-F, so can't be more
specific.
Click to expand...

OK... maybe I'll get ambitious and follow it through.
 
Bob said:
Unfortunately there's no "ignore" feature for the voltage monitoring.
Click to expand...

I have 5 P3B-F boards here, all dead and raided for parts - but AFAIK
the BIOS is similar to the P2Bs, so you should be able to go to the
power management page, scroll to the -12v reading, and hit Page Down to
change it to Ignore.
 
so you should be able to go to the
power management page, scroll to the -12v reading, and hit Page Down to
change it to Ignore.
Click to expand...

You know, some days I'm an idiot. :-)

I don't know what I did the other night, my only excuse is that it was
very late when I hunted down the problem. I switched it to ignore...
I should be all set. We'll see at the next cold boot.

Thanks for poking me with a stick,
Bob
 
Back
Top